International Journal on Engineering Performance-Based Fire Codes, Volume 6, Number 2, p.

53-71, 2004

EXTENDING THE PRINCIPLES OF IMPULSE VENTILATION IN

TUNNELS TO APPLY TO SMOKE CONTROL IN CAR PARKS

H.P. Morgan
International Fire Consultants Ltd, UK

B. Vanhove and J-C. De Smedt

N.V. International Fire Safety Engineering & Technology S.A., Belgium

(Received 23 July 2003; Accepted 24 November 2003)

ABSTRACT

In the past few years, there have been an increasing number of Impulse Systems installed in car parks, especially
in Northern Europe. Unfortunately, no design method has been described openly in the public domain. This
places Authorities Having Jurisdiction (AHJs, or Approving Authorities) in a difficult position when faced with
a proposal. The present paper is intended to develop calculation procedures which can fill this gap, as well as
pointing to aspects needing further research.

The paper describes principles and calculation methods for road tunnels, and generalises these to the different
geometry typical of open-sided car parks. These principles and calculation methods are then modified to apply
to enclosed car park storeys which have an exhaust fan or fans to finally remove smoky gases.

The fundamental characteristic of these methods is that fans (jet fans) are used to accelerate the bulk air below
the hot sub-ceiling smoke layer to a sufficient velocity to prevent the advance of the leading edge of the smoke
layer. The objective is to allow a clear-air safe approach for firefighters from one side of the fire at the cost of a
more general smoke-logging beyond the other side. Different formulae for critical bulk air velocities are
suggested for flat ceilings; for downstand beams forming channels parallel to the fan jets; and for downstand
beams at right angles to the fan jets. A method of relating this critical velocity to the number, size, and emission
velocities of the jet fans is suggested.

The desirability of matching the final exhaust to the induced bulk airflow is indicated for enclosed car park
storeys in order to avoid recirculation of smoke into the area intended to be kept clear.

Finally, the need to employ a careful scaling of key parameters in any hot smoke test intended to demonstrate
performance of the system is indicated.

1. INTRODUCTION more onerous requirements for fire-resistance

1.1 Background
Smoke Control in above-ground car parks is
commonly specified in prescriptive Codes (e.g.
Approved Document B to the Building Regulations
[1] in the UK). The approach is to specify openings
in at least two walls, including two opposite sides
of the building, having a minimum free area of a
specified percentage of the floor area. In the UK
Approved Document B [1] requires a total vent
area in the walls of at least 5% with at least 2½ %
in each of two opposite walls for the car park to be
regarded as open-sided. If the car park is not
regarded as open-sided, it should have at least 2.5%
total vent area with at least 1¼ % on each of two
opposite walls. The latter of the two options attracts
rating for the building’s structure than the former.
We can note that this method depends on the
natural buoyancy of smoke on windless days, and
that wind forces can dominate any buoyant forces
likely to be found.
While it is possible to design Smoke and Heat
Exhaust Ventilation Systems for car parks [e.g. 2],
the practicality of this approach is greatly
influenced by the lack of headroom beneath beams
(or ceilings) commonly found in car parks.
An alternative method of smoke control has been
used in tunnels for many years [3]. This method
adopts the principle of “Impulse” or “Jet Fan”
ventilation.

53
International Journal on Engineering Performance-Based Fire Codes
1.2 Impulse Ventilation in Tunnels:
Qualitative Principles
A fire occurring in a tunnel, for example in a
vehicle which has come to a halt, will cause a
flame and smoke plume to rise. This plume will
reach the roof, and spread outwards in both
directions. If the fire is large enough for flame
temperature gases to reach the roof, the flames then
spreading under the roof cause a rapid increase in
heat radiation falling on the fire and on adjacent
vehicles, which can lead to a rapid acceleration in
growth rate of the fire as it develops and spreads to
other vehicles.
The hot smoke layer beneath the roof advances at
speeds of up to 5 ms-1 for flame temperature gases.
Typical vehicle speeds in a road tunnel would be
above 30 kph. This corresponds to speeds greater
than 5.5 ms-1.
It follows that, in a single-bore tunnel with traffic
travelling in just one direction, vehicles in front of
the vehicle on fire will be able to drive to safety
faster than the hot smoke will advance. Vehicles
behind the fire will be forced to stop, and will then
be at risk from the growing fire (see Fig. 1). In a
tunnel where traffic moves in opposite directions in
the same bore, vehicles travelling in both directions
may become trapped by the fire.
There is an aerodynamic interaction between the
advancing front of the hot smoke layer and the air
in the tunnel. The smoke has to “push aside” the air
in front of it, resulting in a resistance to that
advance. This process is illustrated in Fig. 2. This
Fig. 1: Impulse ventilation system in a tunnel
Fig. 2: The advance of the leading edge (or “nose”) of a buoyant layer of hot gases
54
resistance depends on the difference in velocity
between the hot smoke and the air. If the bulk
velocity of the air in the tunnel is equal to and
opposed to the speed at which the hot layer is able
to advance in still air, the resultant velocity of
advance of the hot smoke will be zero. In other
words, inducing a large enough velocity in the bulk
air flow will prevent the hot smoky gases from
endangering stopped vehicles on the “upstream”
side of the fire – i.e. in the direction the bulk air
comes from. The turbulence expected at the leading
edge (or “nose” if we follow Hinkley’s [4]
nomenclature) of the hot smoke will produce a
great deal of mixing of smoke and air, causing the
tunnel to become smokelogged on the
“downstream” side of the fire.
Where the vehicle direction of travel is aligned
with the bulk air flow this smokelogging will not
affect safety (vehicles will have driven clear).
There will on the other hand be safe conditions for
vehicles trapped by the fire on the “upstream” side,
and clear-visibility access for firefighters
approaching from the “upstream” side.
Smoke control systems using fans to transfer
momentum into the bulk air in the tunnel in order
to achieve the desired velocity to prevent the
advance of hot smoke in the “clean” direction are
known as Impulse Ventilation Systems. The fans
used to introduce this momentum are commonly
known as jet fans, and the concept is often called
“Jet Fan Ventilation” as an alternative to “Impulse
Ventilation”.
 International Journal on Engineering Performance-Based Fire Codes
1.3 Impulse Ventilation in Car Parks:
Qualitative Principles
The principles developed for tunnels can be applied
to the more complicated circumstances of a car
park. In the past few years, there have been an
increasing number of Impulse Systems installed in
car parks, especially in Northern Europe [e.g. 5].
Unfortunately, no design method has been
described openly in the public domain. This places
Authorities Having Jurisdiction (AHJs, or
Approving Authorities) in a difficult position when
faced with a proposal. The present paper is
intended to develop calculation procedures which
can fill this gap.
The paper will start by detailing the design
methods appropriate to tunnels, and will then
generalise these to open-sided car parks. It will
then discuss the further aspects of enclosed car
parks, pointing to differences in approach
compared to open-sided car parks. Finally, there is
a brief discussion of acceptance tests for Impulse
Systems in car parks.
2. DESIGN OBJECTIVES FOR SMOKE
CONTROL IN A CAR PARK STOREY
Smoke control can serve several purposes in a car
park.
It can ensure that escape routes within the car park
storey continue to be usable even while the fire is
burning. This objective is therefore to protect the
Means of Escape (MoE).
In practice, this objective can only be met if escape
routes are kept free of smoke to a great enough
height for people to move freely in clear air. In
most car park designs, this objective is of minor
important compared to the provision of suitably
short travel distances leading to protected stairs
and/or fire exit doors.
Smoke control can provide “smoke clearance”.
This involves clearing smoke out of the storey after
the fire has been suppressed. The objective is to
provide good visibility to help firefighters to
confirm that there are no secondary fires, and also
to speed the reinstatement of the building to normal
activity.
Between these two extremes, smoke control can
assist firefighters in finding and extinguishing the
fire. In other words, it can make operational fire
fighting faster and more effective, thus reducing
damage to the building. This objective does not
require as good a performance as for protecting
occupants and their escape route, in view of the
special equipment and training available to fire
services. Indeed, the smoke control system need
not necessarily be activated until firefighters arrive
on site.
Impulse systems are usually designed to allow
firefighters clear access with good visibility to one
side of the fire, at the cost of creating turbulent
mixing of smoke and air on the other side of the
fire, creating extensive smoke logging. It follows
that it is usually safer if Impulse Systems are not
started while people are evacuating (as they may be
caught in the fan-induced smoke logging quicker
than with the fans off), but are instead started after
a delay comparable to the firefighters attendance
time. In practice, a delay of 5 minutes from
detection is often used.
It may also be noted that many smoke control
systems are dual purpose: they operate
continuously to dilute engine fumes etc. to safe
levels, and change to the (usually) higher
ventilation rate needed for smoke control when a
fire is detected.
A consequence of the previous two paragraphs as
far as Impulse Systems are concerned, is that the jet
fans may need to be turned off when smoke is first
detected, and then turned back on again at the
smoke control level after the pre-determined
interval.
Even though this present paper is based on
similarities between tunnel systems and car park
systems, it is worth recognising that there are many
differences between the two. Some of these are
summarised in Table 1.
3. IMPULSE VENTILATION IN
TUNNELS
3.1 Principles
Design calculations must first establish the largest
fire appropriate for the circumstances. Where the
design is based on a steady-state (i.e. on a “worst
plausible case”) fire, this can be specified as a heat
generation rate coupled with an effective perimeter
of the fire. The mass flow rate and temperature of
smoky gases reaching the roof close above the fire
can then be calculated using established formulae
and the known height of the tunnel.
The rate of advance of the front edge, or “nose”, of
the smoke layer must be assessed. The outcome of
this calculation is a critical velocity of advance of
the hot smoke layer under the ceiling, for the
specified fire in a tunnel of specified height and
width.
55
International Journal on Engineering Performance-Based Fire Codes

Table 1: Some typical differences between tunnels and car parks (Note: Many exceptions exist)

Tunnel Car park

Aspect ratio (transverse section) Height similar to width
Aspect ratio (long section) Very long compared to width
Vehicle movement Unidirectional, at road speeds
Vehicle types Mixed. Cars to trucks and buses
Typical design fire Truck. > 17 MW
Air flow pattern induced by jet Along length of tunnel in
fan direction of traffic flow
Jet fan operation As soon as fire is detected
Evacuation patterns for people If ahead of fire, drive on. If
behind fire, escape on foot
away from fire
Fire fighting approach Drive fire appliance into tunnel
from air inflow end, as close as
possible to the fire
Height << width
Length similar to width
Multi-directional, at slow speeds
Cars
Car. About 3 MW
Specific to each car park.
Ignores traffic movement
Delay until pedestrian escape is
complete
Escape on foot via building’s
Means of Escape (e.g. protected
stairwells)
Enter building on foot, approach
fire following jet fan-induced
airflow

We then need to be able to calculate jet fan
characteristics capable of inducing a bulk airflow at
least equal but opposite to the smoke layer’s speed
of advance in the “upstream” direction. The
transfer of momentum from fan jet to bulk air,
resulting in the creation of a pressure pushing the
air, is discussed below.
This pressure has to be sufficient to overcome the
pressure losses experienced by the bulk air while
travelling through the tunnel. In general, a tunnel
will have an entry and an exit opening to the
atmosphere. There may be bends or changes in
cross-section. There will be friction effects at the
walls, floor, and roof. The tunnel is in effect a
large duct, similar in most respects to an air-
handling duct in a HVAC system in a building.
Design information developed for ducts [6] can be
used to assess the flow resistance (assessed in
terms of pressure losses at each contribution to
flow resistance such as entry losses, exit losses,
losses at bends, etc.) of the tunnel. The design
should be successful when the sum of all the
pressure losses, for induced bulk air speeds greater
than or equal to the critical “nose” velocity, is
equal to the driving pressure induced by the jet fans.
This will ensure that the smoke will not be able to
advance in the “upstream” direction from the fire.
The remainder of Section 3 develops these ideas
quantitatively.
Wind pressure differences between the entry and
the exit to the tunnel can either assist or oppose the
flows induced by the jet fans. Daly [3] suggests
that such effects should be taken into account. This
aspect is not developed further for tunnels in this
present paper, although the effect of wind pressures
is considered below in the context of car park
Impulse Systems.
3.2 Jet Fan Thrust and Induced Velocities
in Tunnels
3.2.1 Thrust and induced pressure differences
The principles of Impulse or Jet fan ventilation in a
single-tube tunnel (or covered roadway) have been
described in Section 3.1 above. The present
purpose is to develop the mathematical formulae
needed to implement a design based on those
principles. Equivalent explanations can also be
found elsewhere [e.g. 3,7].
One or more fans will be arranged so that they emit
jets along the length of the tunnel. These jets
should be arranged such that they minimize contact
with the roof and walls of the tunnel. It is desirable
for the jets to have the maximum opportunity to
interact with the air in the tunnel without friction
effects slowing the jets before that interaction has
occurred.
A fan can be characterized as having an emission
orifice of area Af m2 and an emission velocity of vf
ms-1.
The volume flow rate emitted by the fan is:
V f = Af v f (1)
For ambient temperature air of density ρ0 kgm-3,
the mass flow rate of air emitted per second is the
product of volume flow rate and density, i.e.:
Mf = ρ0 Vf (2)

56
 International Journal on Engineering Performance-Based Fire Codes
The total momentum in the air emitted each second
(the momentum flux) is simply the product of mass
flow rate and velocity, i.e. for a single fan
Momentum flux = Mf vf = ρ0 Vf vf (3)
For a free jet – that is for a jet which does not
interact with resisting solid surfaces such as walls
and roof – the total momentum is conserved despite
the entrainment of air into that jet. That is, the
mass flow rate in the jet increases as air mixes into
the jet from the air surrounding the jet; the velocity
in the jet decreases, but the product remains
constant.
Newton’s Second Law also applies to fluids. The
rate of change of momentum corresponds to a force.
This force, when exerted by a fan is usually called
a thrust. In our case, the jets will slow until they
become part of the bulk air movement in the tunnel.
In other words, the momentum flux emitted per
second by the fans is transferred to the air in the
tunnel. Hence, the total momentum flux in the bulk
air is:
ρ0 At vair2 = ρ0 Vf vf (4)
where At is the cross-sectional area of the tunnel at
a right angle to the direction of flow of the air.
The increase in momentum flux in the bulk air is
experienced as a force (Newton’s Second Law)
acting on that bulk air volume. It is convenient to
recall that a force exerted evenly on a fluid can be
expressed as a pressure. It follows then that the
rise in pressure above ambient in the tunnel due to
one fan is just:
ρ 0 At v air
2 calculate ∆ ploss for a given value of v, via λ and D
∆p = (5)
At
i.e.
ρ 0V f v f
∆p = (6)
At
where there are N fans in the same length of tunnel,
we have
Nρ 0V f v f
∆p = (7)
At
This is the impressed pressure difference driving
the air through the tunnel, against the resistance of
the pressure losses due to bends, surface roughness,
etc. These losses are described in more detail in
Section 3.2.2 below, and will depend on the design
air velocity which is in turn derived from the speed
of the “nose flow” (see below).
It is pointed out by Daly [3] that in practice, the
impressed pressure is based on the difference
between vf and vair. It follows that where greater
accuracy is required, vf in equations (1) to (7)
should be replaced by (vf – vair). In the present
work, this correction is ignored for simplicity, but
it can be noted that it can be introduced as a post-
hoc correction once the value of vair has been
calculated.
3.2.2 Pressure losses in the tunnel
To a good approximation, a tunnel can be regarded
as a large duct. Provided that all significant flows
are turbulent, that is provided that the Reynolds
Numbers are large enough, the difference in scale
becomes unimportant, and the resistance to airflow
in the tunnel can be calculated by the same
methods as in a HVAC or ACMV duct. Methods of
calculating resistance to flow are well known. See
for example Section 4 of ref. 6.
We follow ref. 6 by noting that the energy in fluids
moving through ducts is lost in two major ways.
One is by frictional interaction with the walls. This
can be expressed in D’Arcy’s equation for pressure
loss due to friction:
l ρ 2
∆ ploss = λ v (8)
D 2
where the friction factor λ is a function of
Reynolds Number and the relative roughness of the
wall surfaces. For further details on how to
(the hydraulic mean diameter of the duct/tunnel),
see ref. 6 Section 4.
The other major source of energy loss (i.e. of
pressure loss to overcome resistance to flow)
occurs when a bend, obstacle, or other feature
generates additional turbulence in the fluid flow.
These losses can be expressed in “Pressure Loss
Factors” ς defined by:
ρ
∆ ploss = ς v2 (9)
2
The pressure loss factor is usually determined
empirically, and typically is a shape-dependent
function rather than a scale-dependent function.
Although it may be necessary to extrapolate
(exercising due caution) from an air conditioning
duct to a similar geometry tunnel obstruction, many
values of ς factors can be found in Section 4 of
57
International Journal on Engineering Performance-Based Fire Codes
Ref. 6. One can note particularly that a tunnel will
usually have an entry loss (where air is drawn in
one end) and an exit loss (where air and smoke are
blown out the other end); plus an additional
pressure loss term for each geometrical feature (e.g.
bends, changes of cross-section etc.); plus the
pressure loss due to friction over its length. Note
that where there are significant changes in cross-
section or of surface roughness, it will be necessary
to calculate the sum of the pressure losses due to
friction in each length of tunnel.
None of the thermally-buoyant ceiling jet models
currently available make allowance for sloping
ceilings. Sloping tunnels will affect the entrainment
into an inclined buoyant flow, although where the
leading edge of that flow is halted close to the fire
plume such effects will be minor and can be
ignored. There should be minimal effect on the
thermal plume rising above the fire. A gradient in
the tunnel will have no effect on a non-buoyant
flow. It is assumed throughout this paper that the
smoke and fire gases become so mixed with air by
turbulence on the downstream side of the fire that
they can be treated as essentially non-buoyant. It
follows that changes in gradient typical of
roadways can be ignored.
A simple entry (eg. a hole in a wall at right angles
to the duct or tunnel) will often have ς = 0.5. A
similarly simple exit will have ς = 1.0. Local
topographical features will often change these
values [7] and tunnel designers may even have to
model the surroundings in order to arrive at
empirical values appropriate to their circumstances.
It is especially important to note that v in equation
(9) is the velocity local to that particular obstacle,
and changes in cross-section involve changes in
local velocity. The volume flow rate (ignoring any
heating of the air) will be the same value
everywhere. The velocity in D’Arcy’s formula
(equation 8) also must be altered for every change
in section of the tunnel.
In general, a state of dynamic equilibrium will be
reached where
Nρ o V f v f ρ0 lm ρ 0 2
= ∆p = ∑ ς n vn2 + ∑ λm vm
At all n 2 allm dm 2
(10)
where all air is assumed to be at ambient
temperature; the subscript n denotes separate
obstacles, bends, entry, exit or other shape-
dependent features such as changes of section; and
subscript m denotes individual lengths of tunnel (or
duct) having the same cross-section and surface
roughness.
58
In general,
Vair (11)
v n, m =
A t (n , m)
where Vair is the volume flow rate through the
tunnel; vn,m is the velocity local to the pressure-loss
generating feature identified by subscript n or m,
and At(n,m) is the cross-sectional area of the tunnel
at the pressure-loss generating feature identified by
subscript n or m.
This means that for a known geometry of tunnel,
one can either specify Vair (as for example where a
minimum dilution rate of engine fumes including
carbon monoxide must be achieved) or one can
specify v (as for example when trying to prevent a
thermally buoyant smoke layer travelling against
the direction of the fan jets). Then, via equations
(10) and (11), the minimum fan specification can
be calculated in terms of NVf v f (or NA f v 2 ).
f
This immediately shows us that there is a trade-off
between the size of each individual fan and the
number of fans.
3.3 Opposing the Advance of a Thermally-
buoyant Ceiling Jet of Smoky Gases
3.3.1 General
In Section 3.2, it has been shown that jet fans can
induce a bulk flow in the air in a tunnel, determined
by many factors including the desired air volume
flow rate or air velocity. The designer of an
impulse smoke control system must select an
appropriate value in order to achieve his objective,
which is to ensure that all smoke generated by a
fire travels in one direction away from the fire
(usually in the direction of traffic movement),
leaving the other side of the fire free of smoke.
This protects anyone forced to stop; and it allows
firefighters to approach the fire in safety, in order
to extinguish it. In Section 3.3 of this paper, we
discuss how an appropriate value of bulk air
velocity can be specified.
3.3.2 The design fire
The fire is likely to be in a vehicle. Usually the
designer will be interested in a steady-state design
approach. This involves identifying an appropriate
heat release rate and fire perimeter (or, for point-
source plume models, a fire diameter). From these
the initial parameters can be deduced for the
thermally-buoyant smoke plume flowing under the
ceiling.
Useful sources of data on the size of vehicle fires in
tunnels are relatively sparse. See Ingerson and
Romanov [7] for experimental data including a bus
 International Journal on Engineering Performance-Based Fire Codes
and truck fires done as part of the EUREKA project.
Schleich et al. [8] is a useful source for
experimental data on car fires involving modern
cars (e.g. Fig. 3). It should be noted that the car
design fire cited by Morgan et al. [2] of 3 MW
convective heat flux and 12 metres perimeter
derives ultimately from 1960s data, but is broadly
compatible with Schleich et al.’s results prior to the
time when spread to another car is predicted [8].
Fig. 3: RHR vs Time
(Source: Prof. J.B. Schleich)
Once a design fire appropriate to the circumstances
has been selected, one can use the Large Fire
Plume Model to estimate the mass flow rate of hot
smoky gases reaching the ceiling [2]:
M = C e P Y 1.5 (12)
where M is the mass of smoky gases rising past
height Y, Ce is a constant taking the value 0.19
where the fire plume rises vertically to a high
smoke layer base; or the value 0.21 where the layer
base is close to the fire; or 0.34 where the air
approaches the fire from one side and causes the
plume to “lean” at an angle to the vertical. P is the
fire perimeter and Y is the height of rise.
This development of the “Large Fire Plume Model”
was by Hansell [9] drawing on work by Zukoski et
al. [10] and Quintiere et al. [11] (using “virtual
point-source” axisymmetric plumes) to modify the
Large Fire Plume Model developed originally by
Thomas et al. [12] and Hinkley [13].
As a first approximation, we can take Y to be the
height to the roof above the fire. Ce should take a
value appropriate to a “leaning plume”, as the
purpose is to induce a strong air flow in the tunnel
which will inevitably make the plume lean away
from the vertical. Hence take
C e = 0.34 (13)
The average layer temperature close to the plume is
then
Q (14)
θ=
Mc
If greater accuracy is required, M, Q and θ can be
used to calculate the layer’s depth (see ref. 2, or
Hinkley’s method [4] described below); from
which one can assess a new Y value, and
successive iteration can converge on better values
of M and θ. It is suggested that the level of
precision implied by iteration cannot usually be
justified for car park smoke ventilation in view of
the need to introduce safety margins to allow for
unquantifiable pressure losses.
3.3.3 Ceiling jets: Established flows
The design of an impulse smoke control system
first requires an assessment of the speed of advance
of the ceiling jet travelling outwards from the fire –
in the absence of an imposed airflow.
A ceiling jet is defined in the SFPE Handbook
Section 2-4 [14] as being the relatively rapid gas
flow in a shallow layer beneath the ceiling surface
which is driven by the buoyancy of the hot
combustion products. An older usage reserved the
term for an under-ceiling layer flow which still
carried some kinetic energy from the upward
moving gases in the plume above the fire.
We can identify two characteristically different
forms of flow. One (see Fig. 4) is where the smoke
flow has spread under the ceiling until the gases
have reached a sink and have been removed. In
most experiments, this corresponds to a flow which
has reached the edge of the ceiling and is free to
spill past it. This is an established flow. The second
(see Fig. 5) is where thermally buoyant smoke
layers advance beneath a ceiling by displacing the
air in front of the “nose” of the smoke layer. This
air is effectively pushed downwards by the layer.
The energy required to do this comes from the
smoke layer and so displacing this air can be
regarded as a resistance to the “nose” of the layer
flow – an effect absent in a flowing established
ceiling jet.
The nature of the ceiling jet is also influenced by
the geometry of the ceiling. For the purposes of
this paper, we can consider two simple
circumstances. One is where the buoyant layer (in
the absence of any induced cross-wind) flows
radially outward under a flat ceiling; the other is
where the buoyant layer is confined and channelled
between parallel walls or downstand beams.
59
International Journal on Engineering Performance-Based Fire Codes

Fig. 4: An established smoke layer – Smoke flows above cold air

Fig. 5: An advancing smoke layer – Smoke flow displaces downwards the air
in front of the leading edge

Formulae for the velocity of advance of an the fire plume on the ceiling. If an Impulse System
established radial ceiling jet can be found in the is to be designed to prevent the advance of the
SFPE Handbook [14], having been developed by smoke in one direction, the induced airspeed in the
Alpert [15]. These are opposite direction required to achieve this will be
smaller if the smoke is allowed to travel further
1 from the point of impingement. It follows that
Q 3
(15)
v = 0.96  there can be an implied trade-off between the
h induced bulk airspeed and the distance the smoke is
allowed to travel in the unwanted direction before
for r/h ≤ 0.15, and being halted and turned back. In practical design
terms, the maximum smoke travel distance in the
1 1
0.195Q 3 h 2 unwanted direction can be related to the fire
v= (16) services’ willingness (or otherwise) to tolerate
5
r 6
some turbulently mixed smoke in the closest
approach to the burning vehicle.
for r/h > 0.15.
It can be noted that radial ceiling jets are unlikely
Equations (15) and (16) do not require prior to feature usefully in Impulse system designs for
calculation of the entrainment into the fire plume, tunnels. They may, however, feature prominently
as this is included in the correlation. in designs for car parks having extensive flat areas
of ceiling.
It can be seen from this correlation that the layer
velocity is almost inversely proportional to the Ceiling jet flows under confined or channelled
radial distance from the point of impingement of ceilings are discussed in the SFPE Handbook [14],

60
 International Journal on Engineering Performance-Based Fire Codes
although no actual formulae for the layer’s velocity
are given. Characteristic velocities for established
channelled flows were developed by Morgan and
Hansell [16] and are cited in Morgan et al. [2].
They identified the effect of discharge coefficient
on the velocity as the hot smoky gases flow
beneath downstand beams at right angles to the
flow direction:
Where the channelled flow has no downstands (i.e.
a flat ceiling between the side walls or constraints)
the characteristic layer velocity is [2,16]):
1 “nose” of a layer flow along a tunnel takes the form:
 gQT  3
v = 1.27  (17)
 cρ 0WT0
2
 v nose  ( h − d )( 2h − d ) 
Where there are downstands across the flow
direction:
1 the buoyant layer depth and
 gQT  3
v = 0.76  (18)
 cρ 0WT0
2
  gQT
In a fully channelled flow, equation (17) will apply
over the full length of the channel or tunnel.
Equation (18), on the other hand, only applies to
the flow as it passes beneath the downstand.
In order to apply equations (17) or (18), it is
necessary to pre-calculate the layer temperature.
This can be done using equations (12) to (14).
The depth of the flowing layer (in the absence of a
counterflow of air below) is given by Morgan et al.
(equation 5.11 of ref. 2) as:
2 Fig. 4 (reproduced herein as Fig. 6). His Fig. 5
0.36  MT  3
d=   (19)
C d  θ 0.5WT00.5 
where Cd takes the value 1.0 for a flow under a flat
ceiling between beams W metres apart, and takes
the value of 0.6 for the flow under a deep beam
across the direction of flow.
In general, because all established flow formulae
ignore the resistance to flow at the leading edge or
“nose” of the layer, they are likely to predict a
higher velocity than will actually be the case at the
leading edge of the ceiling jet. In terms of the
design of an Impulse system, this will lead to a
higher jet fan-induced airspeed to oppose the
smoke movement, and the use of such formulae can
therefore be expected to err on the side of safety in
design.
3.3.4 Ceiling jets: Leading edge or “nose” flows
Hinkley [4] has discussed the speed of advance of
the leading edge of a smoke layer (i.e. a ceiling jet)
in a corridor or tunnel, as a part of a study of smoke
movement in shopping malls. His work took
account of the nose of the ceiling flow explicitly,
and so we can expect his work to be the most
appropriate source for the present circumstance.
Unfortunately he only developed formulae for
channelled flows without a downstand across the
flow, which serves to limit the applicability of his
results.
Hinkley [4] has shown (see his equation (16),
modified for higher temperature by setting U’=
u(T/To)1/3) that the velocity of advance of the
1
3
=  (20)
U ' C  h( h + d ) 
where h is the height of the tunnel or channel, d is

1
3
U ' =  

(21)
 cρ o To W
2

where W is the width of the tunnel or channel.
Hinkley developed a graphical solution method
where one first calculates:
P( gh ) 2
1
(22)
U ′WC
(making use of equations (12) to (14) above). The
depth of the flowing layer is given by Hinkley’s
(reproducd here as Fig. 7) gives the dimensionless
layer velocity v/U΄C and hence the actual velocity
of advance of the leading edge (v).
C is an empirical constant found by Hinkley [4] to
be approximately 0.5 based on a Japanese
experiment in a car park having a complicated
geometry. He also speculated that C would tend to
1.0 as the advance of the nose was slowed by an
opposing bulk airflow, due to a reduction in friction
at the ceiling. We should recognize, however, that
flow resistances due to friction at a smooth ceiling
are likely to be smaller than those due to
displacement of air in front of the “nose” – which
would depend on the relative speed of the buoyant
layer to the bulk air. Hence C is more likely to be
near the empirical value even when the “nose” has
zero net speed relative to the ceiling.
We can note that he also cited Benjamin as having
derived a value for C of 0.82. In practice a higher
value for C will lead to a higher value for the
critical speed of advance of the “nose” of the
61
International Journal on Engineering Performance-Based Fire Codes
smoke layer, and consequently to a larger jet fan
specification, i.e. it will err on the side of safety.
It seems reasonable to adopt Benjamin’s value for
C for design purposes where Hinkley’s method is
adopted.
Hinkley’s analysis was for a smoke layer traveling
in one direction only away from the fire. Where
smoke travels in both directions away from the fire,
it is reasonable to take W as being twice the actual
width, as the smoke flow is equivalent to a one-
direction flow in a tunnel of twice the width.
Fig. 6: Dimensionless layer depth vs
dimensionless flow rate, from Hinkley [4]
Fig. 7: Dimensionless velocity vs dimensionless
flow rate, from Hinkley [4]
62
3.4 Suggested Design Procedure
a) Identify an appropriate design fire for the
circumstances.
b) Estimate separately for each length of tunnel
having the same width, ceiling height and
surface roughness, the layer temperature at
the ceiling using Section 3.3.2.
c) Identify the ceiling jet formula appropriate to
the circumstances for each separately
identifiable length of the tunnel.
d) Calculate the layer velocity for each such
length using Section 3.3.3 or 3.3.4. This will
be the minimum value of induced velocity
acceptable in that length, and can be
designated the critical velocity for that length.
e) Calculate the volume flow rate for each length.
Take the maximum value as the critical
volume flow rate. Recalculate the local
velocities corresponding to all lengths and
obstacles.
f) Calculate the total pressure losses in the
tunnel using these local velocities and Section
3.2.2.
g) Equate this total pressure loss value to the
induced pressure due to the jet fans, and
calculate the overall fan specification in terms
of NVfvf for the tunnel.
3.5 Some Practical Considerations
The positioning of jet fans should be such that the
jets can transfer momentum to the air without
playing on bends or obstacles. The jets should be
able to transfer momentum to the air in the tunnel –
which implies that a jet fan can be close to an entry
blowing inwards, but should never be close to an
exit blowing out. It also seems reasonable to have
a sufficient horizontal separation along the tunnel
for momentum to be transferred before the jet
interacts with the next fan downstream. In order to
maximise the effect on the smoke layer, the ideal
jet fan jet should be angled downwards to avoid
contact with the ceiling, and should avoid contact
with the walls.
4. IMPULSE VENTILATION IN OPEN-
SIDED CAR PARKS
4.1 General
A car park storey can be thought of as a tunnel
which is very wide but (usually) with a low ceiling.
Where there are openings at opposite sides of the
storey, the principles of Impulse ventilation can be
applied with little modification – in principle (See
Fig. 8). The most important differences between a
 International Journal on Engineering Performance-Based Fire Codes
tunnel and a car park storey arise from the much
greater horizontal extent of the storey.
4.2 Design Objectives
Vehicles cannot drive as quickly in a car park as in
a road tunnel. There is therefore essentially no
opportunity to drive away from the smoke faster
than it moves. It follows that Impulse systems in
car parks cannot protect escape with the cars. As
explained above, they cannot protect the MoE for
occupants in view of the turbulent mixing in the
direction of the fan jets, and so there is a need for
evacuation to be accomplished using travel
distances and protected escape stairs etc., before
the jet fans are switched on.
The primary role of impulse systems in this context
is to provide clear access from one side of the fire
for firefighters, so that they can control the fire.
The system will then provide smoke clearance so
that the building can be put back into use as early
as possible.
Fig. 8: Impulse ventilation in an open car park
4.3 Opposing Thermally-Buoyant Ceiling
Jets
The principles and formulae involved are the same
as in Section 3.3 above. The greater horizontal
extent of the car park compared to a tunnel, and the
lower ceiling, increases the importance of radial
ceiling jets (where there is a flat ceiling), and
increases the importance of downstand beams
where they are present (discussed further in Section
4.4 below). Design of an impulse system can
follow a similar sequence to that for tunnels. An
appropriate design fire should be selected. This is
commonly based on a single car (e.g. 4.5 MW heat
release rate, 12 m perimeter cited in ref. 2), with
the implicit or explicit assumption that the fire
service would be able to intervene before the fire
spreads to neighbouring cars (typically this
assumes that the firefighters can intervene in the
fire in the first 10 to 12 minutes after ignition (see
for example Schleich et al. [8]). If attack times are
likely to be longer, then perhaps consideration
should be given to fitting sprinklers, or assuming a
larger design fire.
63
International Journal on Engineering Performance-Based Fire Codes
The selection of a suitable formula for the advance
of the ceiling jet is similar to that for a tunnel
discussed in Sections 3.3.3 and 3.3.4 above. Where
the ceiling is flat, the ceiling jet will be radial, and
equations (15) and (16) will apply. Note that the
trade-off between an acceptable degree of
smokelogging close to the fire on the best line of
approach for the firefighters, and the critical
velocity of bulk airflow can become a crucial factor
in the design. It is recommended that in all such
circumstances there should be consultation with the
fire service before progressing the design.
Where there are relatively deep downstand beams
parallel to the jet fan jet directions, either equation
(17) or Hinkley’s method (Section 3.3.4 above)
will apply when assessing the critical velocity to
halt the advance of the smoke layer.
The depth of the channelled flow should be
calculated (see Sections 3.3.3 and 3.3.4) to confirm
whether or not smoke will spill into neighbouring
channels. If it will, then the critical velocity
becomes less accurate, although the “non-spillage”
value can still be adopted as being conservatively
safe in the context of calculating jet fan
specifications.
Where there are beams at a right angle to the jet fan
direction, they will act as weirs, spreading the
smoke flow laterally until smoke spills across most
or all of the available length of the beam. This may
be the full width of the car park. This increase in W
in equation (19), coupled with the smaller constant
for transverse beams, can mean that the critical
velocity is much lower for transverse beams than
for parallel beams, leading to a smaller
specification for the jet fans in the former case.
Smoke will tend to spread more widely across the
jet fan’s direction with these transverse beams,
affecting more of the car park storey.
Some car parks have downstand beams in a “criss-
cross” pattern. These cannot be calculated by zone
model methods, but can be expected to perform
somewhere between the parallel and the transverse
beam cases. It may be necessary to use more
geometry-specific methods based on
Computational Fluid Dynamics (CFD) to confirm
the performance of Impulse Systems with such
ceilings.
It should be noted that because the critical velocity,
and thus the specified value of NVfvf for the car
park storey, is different for the different ceiling
configurations, we should not expect designs
shown to work acceptably for one ceiling
configuration to be as satisfactory for another.
64
4.4 Balancing Induced Pressure
Differences Against Flow Resistances
One can identify an entry pressure loss and an exit
pressure loss. In general, the car park storey will
be short compared to a typical road tunnel, and the
cross sectional area of the car park will usually be
large. This means that one could usually ignore the
friction losses and base calculations on the shape-
dependent pressure losses.
The method described in Section 3.4 above can
then be used to calculate NVfvf for the storey.
Where there are many uncertainties in the
calculation (see for example Section 4.5 below), it
can be necessary first to develop the zone model
methods set out in this paper as an initial proposal,
and then use that proposal as a starting point for a
more detailed CFD analysis which would either
validate the proposal, or which would suggest
amendments needed to make the proposal viable.
4.5 Some Practical Design Aspects
The greater width of a car park storey relative to its
height, when compared with a typical tunnel,
makes it more important that the calculated number
of fans can exert a uniform pressure on the air
across the full width of the storey. If this is not
done, the air will tend to move from the region of
greater pressure towards the lower, and there will
be a circulation pattern of air (and smoke being
carried with the air) with some smoke being carried
in the undesired direction past the fire into the areas
intended to be clear. It may be necessary to specify
a larger number of smaller fans to be able to
arrange for the spreading jets to be touching the jets
to both sides and also the side walls by the distance
from the fans where the jet velocities have reduced
to the induced air velocity.
This may mean that all N fans are used to form one
line across the storey, although the non-uniformity
across a typical jet implies that it would be better to
arrange fans in two lines, with one line covering
the “gaps” between fans in the other line (see Fig.
9).
In many car parks there are columns, enclosed
stairwells, or other constructional features which
will generate turbulence in the air flow, but for
which there are no known values of ζ. It is also
unfortunately the case that there are no ζ values in
the available literature for the pressure losses due
to air flowing past, around, and under parked cars.
Indeed, most current designs of impulse systems
appear to ignore the presence of parked cars
altogether, and design essentially for an empty car
park. There is another practical difficulty where
jets play directly onto fixed objects, or onto parked
 International Journal on Engineering Performance-Based Fire Codes
cars, and some of the momentum in the jet is lost
by conversion back into a force on the objects and
is therefore not available for raising the induced
pressure acting on the bulk airflow.
This implies that there are major sources of
pressure loss which are unquantifiable in most
cases. These can be compensated for by increasing
the number of fans. It appears usually necessary in
such cases to adopt a subjectively assessed safety
margin – e.g. by doubling the number of lines of jet
fans calculated as being necessary based on the
known pressure loss terms.
Where downstand beams form channels, there is a
tendency for both the ceiling jet (smoke) and the
fan jet to be attracted to the ceiling by the Coanda
effect (i.e. where the static pressure is reduced
because of the gas velocity – a demonstration of
Bernouilli’s Equation). This may interfere with the
transfer of momentum from jet to air to allow air to
recirculate past the fans in channels between the
fans. It may be necessary to mount a fan in each
channel – although this is often not practical. More
research is needed to study this effect and to
identify whether the problem exists and if so how it
can best be overcome.
If the induced velocity at the inlet opening is too
high, the incoming air jet will entrain air inside the
car park. This can cause a recirculation pattern
which may bring smoke past the jet fans into the
Fig. 9: Staggered implementation of jet fan lines
area we wish to be kept clear. Similar concerns
affect Smoke and Heat Exhaust Ventilation
Systems, for which the recommendation for air
inflows close to smoke [2] is that the incoming jet
should not exceed 1 ms-1. Research into the
significance of this effect would be desirable.
Wind pressures on the inlet and exit openings can
be much larger than the pressures induced by the
fans. A commonly recommended approach is to
use reversible jet fans, with the direction of their
operation linked to wind direction sensors such that
the wind-induced airflow in the car park is always
in the same direction as the fan jets.
Jet fans can be used additional to the calculated
lines, to direct or “steer” flows of air and smoke
within the car park. The purpose is to make bulk
air flows more homogeneous even where the car
park geometry is complicated. It would usually be
necessary to use a CFD model to confirm the
effectiveness of the proposal.
CFD modellers should note the importance of
properly modelling the structure of the jets issuing
from jet fans. Suppliers should note the importance
of making such details available to modellers.
Simply assuming that the jet from a fan is one-
dimensional risks being very misleading. CFD
modellers should also note the critical importance
of mesh size when modelling flows around beams.
65
International Journal on Engineering Performance-Based Fire Codes
Because the impulse system effectively guarantees
smokelogging in the down stream direction, it is
important that the system should be fully integrated
with other fire protection measures. It has already
been noted that systems should not be initiated
until a safe time after detection which will allow
people to have evacuated from the car park. This
carries an implied need for a good smoke detection
system to give warning as early as possible. It also
implies that the alerting process be effective, and
compatible with people perhaps being inside well-
soundproofed cars. This suggests the need for eye-
catching flashing lights to draw attention to written
messages. In view of the difficulty in driving to
safety, the messages should advise people to
evacuate on foot by the nearest fire exit, leaving
their cars where they are.
Protected evacuation stairwells which connect to
the smokelogged areas of the storey risk permitting
smoke to enter the stairshaft, perhaps with the
leakage rate increased by the local pressure rise due
to a fan jet directed towards the stairwell door.
Even though evacuation should be complete before
the jet fans start, and even though firefighters
should be directed towards a clear area of the car
park storey as a part of their approach to the fire,
caution suggests that doors from the car park to the
protected shaft should be fitted with smoke seals to
minimise any such leakage.
5. IMPULSE VENTILATION IN
ENCLOSED CAR PARKS
5.1 Comparison with Open-Sided Car
Parks
The key difference between an open-sided car park
and an enclosed car park is that the air inlet and air
Fig. 10: Impulse ventilation in an enclosed car park with exhaust fans
66
exit openings in the former are natural openings;
whereas the exhaust from the latter takes the form
of an exhaust fan or fans (see Fig. 10 and compare
with Fig. 8). It is usual even in enclosed car parks
for some or all of the incoming air to enter the car
park through natural openings, although systems
can also be designed with fans supplying air as well
as exhausting air and smoke.
The fan exhaust has the effect of making the
system much less sensitive to wind pressures, and
reduces or eliminates the need for the system to be
designed to be bidirectional depending on wind
direction.
The exhaust rate is determined by the capacity of
the exhaust fans. Conservation of mass means that
the air entering the inlets has an equal mass flow
rate to the exhaust. If we ignore the heating due to
a fire, this means that the volume flow rate entering
the car park storey is equal to the exhaust rate.
This also implies that the net volume flow rate
through the car park storey has this same value –
regardless of what any jet fans are doing inside the
storey. In practice, this means that the bulk air
velocity needed to halt the advance of the
thermally-buoyant ceiling jet beneath the storey’s
accent is determined by the exhaust fans, while the
jet fans serve to “steer” the air flows within the
storey in order to achieve a more uniform flow and
to achieve the same effect on the ceiling jet
wherever the fire occurs across the width of the
storey.
The effect of downstand beams, and of excessive
airspeeds at air inlets, are the same as for the open-
sided car park storey.
 International Journal on Engineering Performance-Based Fire Codes
It is still necessary to estimate the number of jet
fans required for the car storey. The pressure loss
at the storey inlet is more appropriately considered
as a part of the exhaust fan’s exhaust duct design
calculation, and need not be considered further
herein. Similarly, the pressure losses at the exhaust
fans’ intakes can be considered as a part of the loss
calculation for the exhaust duct.
The calculation of the critical ceiling jet velocity,
and hence of the critical bulk air flow rate, is the
same as detailed in Sections 3 and 4 above. The
relation between the jet fan specification NVfvf and
the momentum in the bulk air flow (equation (4)
above) can be adopted as a first approximation to
the jet fan specification. This jet fan specification
can be expressed as one or more lines of fans
across the desired direction of smoke movement, as
discussed in Section 4 above. There will be many
unknown sources of pressure loss, especially when
there are vehicles present in the car park, and as in
Section 4 it is desirable to design at least one more
line of fans than calculations indicate. This inherent
uncertainty in these zone model methods makes it
very important that the design should be validated
by a thorough CFD model of the storey, and
perhaps amended in the light of the results. Note
here that the primary role of the initial zone model-
based design is to provide a reasonably-plausible
basis for initial discussions, and to reduce the
number, duration, and hence cost of the CFD runs
required.
5.2 Some Practical Considerations
If the induced bulk air volume flow rate in the
storey is much smaller than the exhaust fan/s
volume flow rate, one expects the exhaust to
dominate the pattern of air movement within the
storey. The effect would be to reduce the
effectiveness of the jet fans in creating a uniform
flow along the storey. This makes it more likely
that smoke can move in the undesired direction
where the induced air velocity is locally less than
the critical velocity of the leading edge of the
ceiling jet. The positive aspect of this is that if the
inlets and the exhausts are evenly distributed across
opposite walls of a rectangular-plan storey, and the
volume flow rate is greater than the critical value,
the benefits of impulse ventilation can be obtained
with no jet fans at all! Unfortunately, few enclosed
storeys have such an ideal geometry.
If the induced bulk air volume flow rate is greater
than the exhaust volume flow rate, the discrepancy
between what is being “pushed” towards the
exhaust and what is being removed by the exhaust
must somehow travel back past the fans to become
available at the fan inlets. This can either take the
form of a recirculation pattern throughout the
storey, causing smoke to affect the areas intended
to be kept clear, or it can take the form of a local
recirculation at each fan, which would have less of
an adverse effect on the performance of the
impulse system. The significance of these
recirculation patterns cannot be assessed by zone-
model methods, but should be revealed by CFD
modelling.
The ideal design will have the bulk airflow induced
within the storey equal to the exhaust volume flow
rate, in which case there should be no opportunity
for adverse recirculation. One can note, however,
that the balance point between induced and exhaust
volume flow rates may be changed by changing the
numbers of cars in the storey. Research into the
significance of this effect would be highly desirable.
The importance of avoiding excessively fast jets of
air entering the storey through inlets of small area
is essentially the same as discussed in Section 4
above. The adverse effect, as in Section 4, is that
entrainment into the air jet inside the inlet can itself
drive a recirculation of smoky air which can
overcome the “steering” effect of the jet fans.
Many enclosed car parks are multi-storey, with
vehicle access ramps connecting the storeys. These
ramps can often result in a complicated pattern of
air flows. This can often be used to advantage in
providing make-up air naturally into the fire-storey,
although success in integrating this into the overall
design depends strongly on the actual geometry of
the building.
It is crucially important that the spread of smoke
through the car park be restricted; in particular it is
important that smoke should be prevented from
spreading through the vehicle access ramps from
one storey to another. The possibility of a fire in a
vehicle on one of the ramps must also be
considered.
Preventing smoke spread between storeys needs to
be a primary design objective of the Impulse
Ventilation System. In many circumstances, it may
prove a more convenient option to use a Smoke and
Heat Exhaust Ventilation System (SHEVS) concept
to contain buoyant smoke and prevent it moving
into the access ramp area. The design of such a
SHEVS is outside the scope of this present paper:
we can however note that in a large and complex
car park, it may be appropriate to mix several
different design concepts in different areas
provided that they are complementary and do not
interfere adversely with each other.
67
International Journal on Engineering Performance-Based Fire Codes
6. ACCEPTANCE TESTS OF
INSTALLED IMPULSE VENTILA-
TION SYSTEMS
Where a full test of the installed system is required
to confirm the satisfactory operation of an impulse
system, one can either do an essentially “cold” test
to confirm that all items of the system perform to
their individual specifications (which unfortunately
does not confirm the performance of the system
when there is an actual fire), or one can conduct a
Hot Smoke Test. The latter can either burn a car-
sized fire, with the inevitable additional repair and
cleaning costs for the storey, or one can use a
scaled Hot Smoke Test. The use of scaling
relationships to create the same observed
performance as an actual full-size design fire but
without damaging the building, has been developed
by Morgan and De Smedt [17]. The conclusions
from their paper can be summarised as follows:
a) Without adopting a fire tray pattern which
corresponds to the scaling relationships in the
gas flows, the observed smoke depths and/or
distances travelled by smoke will not
correspond to the design predictions.
b) Where fans are present, their exhaust velocities
need to be altered to conform to the scaling
relationships in the gas flows, otherwise the
observed smoke depths and/or distances
travelled by smoke will not correspond to the
design predictions.
c) Unless scaling is taken into account when
specifying the hot smoke test fire and any fans
involved, the full design performance of a
smoke control system cannot be properly
assessed.
d) The exception to a) to c) above is when the
same calculation procedure has been used for
both the design condition and for the hot
smoke test, and is fully appropriate to both
conditions. In this case, a good match between
prediction and observation for the hot smoke
test gives confidence in the application of the
calculation procedure to the design condition.
NOMENCLATURE
Af cross-sectional area of emitting orifice of a
fan, m2
At cross-sectional area of a tunnel (or, by
analogy, of a car park storey), m2
c specific heat of air at constant pressure, kW
kg-1K-1
C an empirical constant used in Hinkley’s
method for “nose” flows
Cd a discharge coefficient
68
Ce the dimensioned entrainment constant in the
Large Fire Plume Model, kg m-5/2s-1
d depth below ceiling of a thermally-buoyant
smoke layer, m
D hydraulic mean depth (i.e. effective diameter
of a non-circular duct or tunnel), m
g acceleration due to gravity, ms-2
l length of tunnel having similar cross-section
and smoothness, m
m the subscript m denotes individual lengths of
tunnel (or duct) having the same cross-
section and surface roughness
M mass flow rate of gases, kg s-1
Mf mass flow rate of air emitted from the fan
orifice, Kgs-1
n the subscript n denotes separate obstacles,
bends, entry, exit or other shape-dependent
features such as changes of section
N number of fans predicted as the minimum
necessary for a tunnel or car park storey
P perimeter of fire, m
Q convective heat flow in the gases due to the
fire, kW
r radial distance from the centre of the
axisymmetric plume impinging on a flat
ceiling, m
T absolute temperature of smoky gases, K
T0 absolute ambient temperature, K
u velocity of leading edge of a “nose” flow
(Hinkley), ms-1
U′ leading edge velocity corrected by Hinkley
to allow for higher gas temperatures, ms-1
v gas velocity, ms-1
v air velocity of air in the tunnel or car park
storey, ms-1
vf velocity of air emitted by a fan, ms-1
Vair volume flow rate of air in a tunnel or car
park storey, m3s-1
Vf volume flow rate of air emitted by a fan,
m3s-1
W width of a flowing smoke layer, measured
normal to that flow, m
Y height from the base of the fire to the base of
the hot smoke layer, m
∆p a pressure difference (e.g. impressed by the
momentum of a fan jet), Pa
∆ploss a pressure loss due to either friction or
turbulent energy losses in the air flow, Pa
ζ a zeta-factor: a shape-dependent factor
relating pressure losses to flow velocity
θ temperature of gases above ambient, K
λ friction coefficient
ρ a gas density, kgm-3
ρ0 density of ambient air, kgm-3
 International Journal on Engineering Performance-Based Fire Codes
REFERENCES
1. The Building Regulations 2000, Fire safety,
Approved Document B, Department of the
Environment, Transport and Regions, London,
(2002).
2. H.P. Morgan, B.K. Ghosh, G. Garrad, R.
Pamlitschka, J-C. De Smedt and L.R. Schoonbaert,
“Design methodologies for smoke and heat exhaust
ventilation”, BR 368, CRC, London (1999).
3. B.B. Daly, Woods practical guide to fan
engineering, Chapter 13 Ventilation of tunnels and
mines, 6th edition, Woods of Colchester, Colchester,
UK (1992).
4. P.L Hinkley, “The flow of hot gases along an
enclosed shopping mall - A tentative theory”, Fire
Research Note 807, FRS (Now part of BRE,
Garston, UK) (1970).
5. R. van Beek and G. Dons, “No smoking”, Fire
Prevention and Fire Engineers Journal, Vol. 63, pp.
29-31 (2003).
6. CIBSE Guide C, Reference data, Section 4 - Flow
of fluids in pipes and ducts, The Chartered Institute
of Building Services Engineers, London (2001).
7. H Ingason and L Romanov, “Use of mobile fans in
tunnels”, SP Swedish National Testing and
Research Institute, SP Report 2002:06 (2002).
8. J.B. Schleich, L.G. Cajot and D. Joyeux, “Natural
fires in closed car parks”, Eurofire 99, Brussels,
(1999).
9. G.O. Hansell, “Heat and mass transfer process
affecting smoke control in atrium buildings”, PhD
Thesis, South Bank University, London (1993).
10. E.E. Zukoski, T. Kubota and B. Cetegan,
“Entrainment in fire plumes”, Fire Safety Journal,
Vol. 3, pp. 107 (1981).
11. J.G. Quintiere, W.J. Rinkinen and W.W. Jones,
“The effects of room openings on fire plume
entrainment”, Combustion Science and
Technology, Vol. 26, No. 5-6, pp. 193-201 (1981).
12. P.H. Thomas, P.L. Hinkley, C.R. Theobald and
D.L. Simms, “Investigation into the flow of hot
gases in roof venting”, Fire Research Technical
Paper No. 7, London, The Stationary Office (1963).
13. P.L. Hinkley, “Rates of production of hot gases in
roof venting experiments”, Fire Safety Journal,
Vol. 10, pp. 57-65 (1986).
14. D.D. Evans, SFPE Handbook of Fire Protection
Engineering, 2nd edition, Section 2, Chapter 4 -
Ceiling jet flows, The National Fire Protection
Association, USA (1995).
15. R.L. Alpert, “Calculation of response time of
ceiling-mounted fire detectors”, Fire Technology,
Vol. 8, pp. 181-195 (1972).
16. H.P. Morgan and G.O. Hansell, “Atrium buildings:
Calculating smoke flows in atria for smoke control
design”, Fire Safety Journal, Vol. 12, No. 1, pp. 9-
35 (1987).
17. H.P. Morgan and J-C. De Smedt, “Hot smoke tests:
Testing the design performance of smoke and heat
ventilation systems and of impulse systems”,
FireAsia 2003 “A safe city in motion”, Hong Kong,
26-28 February (2003).
WORKED EXAMPLE
In this example, we look at a simple rectangular car
park storey with a flat ceiling. The dimensions are
not intended to represent a fully realistic design,
but are solely intended to illustrate the principles.
The car park storey is 50 m wide, 75 m long and
has a height of 3.20 m. Both of the shorter sides
are open for the full height, through the full width.
This implies a typical entry coefficient of 0.5, and a
typical exit coefficient of 1.0 by analogy with duct
flows.
The objective is to design an impulse ventilation
system for this car park storey. The design fire has
a convective heat flux of 6 MW and a perimeter of
12 m. This represents a very severe single car fire.
For a real case, the design fire has to be chosen
very carefully in correspondence to the given
circumstances. While dimensioning the jet fan
installation, we shall assume that this installation
must stop the spread of smoke within 15 m on the
“upstream” side of the design fire. The car park
has a flat ceiling which implies a radial spread of
the ceiling jet.
In the calculation, the following procedure is
applied:
Stage 1: Calculation of r/h
r is the radial distance from the centre of the
axisymmetric plume impinging on a flat ceiling,
and h is the height of the car parking.
It can be seen from equations (15) and (16) that we
need to identify whether
r
≥ 0.15
h
For the present example,
r = 15 m
h = 3.20 m
From which we obtain:
r/h = 4.7 > 0.15
69
International Journal on Engineering Performance-Based Fire Codes

Example figure: “Chequerboard pattern” jet fans in an open-sided car park

Stage 2: Calculation of the velocity of the ceiling Stage 4: Calculation of the critical volume flow
jet in the car park storey vceiling jet (ms-1) rate through the storey (and the velocity at the
openings)
Because r/h > 0.15 => we must use equation (16):
The critical volume flow rate is just the product of
1 1 the calculated critical velocity (Stage 2 above) and
0.195Q 3 h 2
the cross-sectional area (Stage 3 above).
ν cj = 5
r 6

i.e. Vair = vceiling jet At

For the present example,
For this example,
Q = 6 MW
h = 3.20 m Vair = 0.66 x 160 = 105.6 m3s-1
r = 15 m
Stage 5: Specification of the characteristics of
From which we obtain: the jet fans

vceiling jet = 0.66 ms-1 In practice, the designer must choose between
available fan types and sizes. For this example,
Stage 3: Calculation of the cross-sectional area assume that fans are available with the following
of the car park storey At (m²) characteristics (Note: It should not be assumed that
actually available fans will have these
This means the cross-section at a right angle to the characteristic parameters):
direction of the induced bulk air flow:
Vf is the volume flow rate of air emitted by a fan =
At = bh 1.4 m³s-1, and vf is the velocity of air emitted by a
fan = 14 ms-1.
For the present example,
Stage 6: Calculation of the number of jet fans N
b = 50 m
h = 3.20 m We need to implement equation (10). The local
velocity in the inlet opening is equal to that within
From which we obtain: the storey, because the opening is full height. The
same is true of the exhaust opening. Therefore the
At = 160 m² local velocity at the “obstacle” represented by each
opening is:

70
 International Journal on Engineering Performance-Based Fire Codes

Vn = 0.66 ms-1

Provided we can ignore the friction losses, equation

(10) then reduces to:

Nρ o V f v f ρ0 ρ0
= ς inlet v n2,inlet + ζ exhaust v n2,exhaust
At 2 2

i.e.

N1.2 ×1.4 ×14 1 .2 1 .2

= 1.0 × × 0.66 2 + 0.5 × × 0.66 2
160 2 2

i.e. N = 2.7

In practice, one would need to use three jet fans, or

find a different model with more suitable
parameters. In general, increasing the total fan
impulse will increase the safety margins within the
storey.

Further design factors

We can conclude that a total of three jet fans will

create a bulk air movement sufficient to achieve
our objective of stopping the advance of the ceiling
jet of smoky gases by a distance of 15 m upstream
of the central axis of the fire plume. These fans
however, need to be arranged abreast to prevent
smoky gases recirculating around the sides of the
jets. Pressure losses due to friction (and the
presence of cars) are not known and have not been
taken into account. There is also the problem that
each fan’s jet is faster in the centre than at the
edges of the jet, so that a fire between fan jets,
closer than the distance needed for full transfer of
momentum to the bulk air, may travel further than
intended before being stopped. This can be
overcome by putting in an extra row of jet fans,
positioned to cover the gaps between the first line.
See the figure in this example which illustrates the
“chequerboard” pattern that results.

If in reality this extra row is more than needed to

overcome the pressure losses, we note that it
increases the bulk air flow and is thus a
conservative amendment to the design.

________________________
Reproduced by permission of International Fire Consultants Ltd. (IFC) and of N.V. International Fire Safety Engineering
and Technology S.A. (IFSET).

71